Professional Documents
Culture Documents
10.1.1.890.6005 Brand Credibility
10.1.1.890.6005 Brand Credibility
of Brand Credibility
and Brand Prestige in
Consumer Brand Choice
Tae Hyun Baek and Jooyoung Kim
University of Georgia
Jay Hyunjae Yu
Sogang University, South Korea
ABSTRACT
This study explores how brand credibility and brand prestige affect
brand purchase intention and empirically investigates how the com-
binatory mechanism of brand credibility and brand prestige materi-
alize across multiple product categories. The proposed model of six
latent constructs is tested with structural equation modeling analy-
sis: brand credibility, brand prestige, perceived quality, information
costs saved, perceived risk, and brand purchase intention. The
results suggest that both brand credibility and brand prestige posi-
tively influence brand purchase intention through perceived quality,
information costs saved, and perceived risk under different product
categories representing the high and low self-expressive nature.
Several implications for advertising messages and brand positioning
strategies are discussed. © 2010 Wiley Periodicals, Inc.
Credibility and perceived value in the brand are often regarded as two impor-
tant virtues marketers need to pursue to build a strong brand (Vanrenen, 2005).
Credibility of a brand, defined as the perceived believability of whether a brand has
the ability and willingness to continuously deliver what has been promised
(Erdem & Swait, 2004), provides unbeatable benefits to both consumers and
H2: Brand credibility will positively affect the information costs saved.
Brand Prestige
As defined earlier, brand prestige can represent the relatively high status of prod-
uct positioning associated with a brand (Steenkamp, Batra, & Alden, 2003; Truong,
McColl, & Kitchen, 2009). An inherent, unique know-how, which concerns a
specific attribute or the overall quality and performance of the product, is the key
criterion for a brand to be judged prestigious (e.g., Dubois & Czellar, 2002). Further,
a higher price (e.g., Lichtenstein, Ridgway, & Netemeyer, 1993; Wiedmann, Hennigs,
& Siebels, 2009; Truong, McColl, & Kitchen, 2009) and the influence of reference
groups (e.g., Bearden & Etzel, 1982) on the consumption of prestige or luxury
brands are often used as proxies for brand prestige, even though they are not
equivalent to brand prestige. Alden, Steenkamp, and Batra (1999) argue that
consumers tend to perceive the consumption of prestige brands as a signal of
social status, wealth, or power since prestige brands are infrequently purchased
and are strongly linked to an individual’s self-concept and social image. Nev-
ertheless, brand prestige does not affect all individuals to the same degree. In
other words, consumption of prestige brands may vary according to the sus-
ceptibility to others (Vigneron & Johnson, 1999). Simply put, publicly self-
conscious individuals are particularly concerned about how they appear to others
and might be more likely to purchase prestige brands. Conversely, privately
self-conscious people are more focused on their inner thoughts and feelings, and
thus would be less likely to purchase prestige brands (Fenigstein, Scheier, &
Buss, 1975). According to O’Cass and Frost (2002), prestige brands differ from
non-prestige brands in some ways that may affect consumers’ purchasing motives
to improve their social standing and self-expression. Compared with non-
prestige brands, prestige brands not only provide intangible benefits to con-
sumers, but also create value for the consumer through status and conspicuous
consumption. Along this line, some scholars have suggested that consumers
may associate global brands as having higher prestige because of their relative
scarcity and higher price compared with local brands (Bearden & Etzel, 1982;
Batra et al., 2000). Wong and Zhou (2005) found that perceived brand prestige
has a greater effect on purchase intention when the product category is of high
social display value.
Brand prestige appears to be linked directly with perceived quality. For exam-
ple, Steenkamp, Batra, and Alden (2003) found that consumer value of a global
brand is positively associated with both brand prestige and perceived brand
quality. Furthermore, their results indicated that brand prestige has a signifi-
cant impact on purchase intention. Vigneron and Johnson (1999) synthesized the
H5: Brand prestige will positively affect the information costs saved.
H7: For the high-expressive product, brand prestige’s total impact on pur-
chase intention will be greater than brand credibility’s total impact.
H8: For the low-expressive product, brand credibility’s total impact on purchase
intention will be greater than brand prestige’s total impact.
METHOD
Procedure
Prior to testing the hypotheses, a pretest was performed to identify product cat-
egories reflecting the high and low self-expressive characteristics and to pre-check
the manipulation of self-expressive products. Details about the specific product
selections and the self-expressive manipulations are discussed later. Samples
for the pretest and the main study were undergraduate college students.
Although the sampling criteria may be a potential limitation to the general-
ization of the results, student samples are one of the most homogeneous segments
of consumers and are among the important target groups for many product cat-
egories. Erdem and Swait (1998) have confirmed the validity of their brand
credibility model using undergraduate student samples.
In the main study, subjects were randomly assigned to a questionnaire that cov-
ered two of the four product classes representing the high and low self-expressive
natures. The data were collected through an online survey-based procedure,
which ensured method invariance. A random link generator was used to ran-
domize questionnaire distribution. Preceding this, analysis of the main study
consisted of two steps. First, the pooled data of all manipulation conditions were
analyzed to test the proposed model using structural equation modeling (SEM)
via the use of LISREL 8.72 (Jöreskog & Sörbom, 1996). Second, multigroup
analyses were conducted to test the moderating roles of self-expressive product
classification.
Pretest Results
In an attempt to identify the high and low self-expressive product categories, an
initial pool of 12 product categories was selected through focus groups. Forty
undergraduate college students (male 40%, female 60%, and age M 20)
were asked to evaluate the self-expressive characteristics of 12 product cate-
gories. With a 7-point Likert scale anchored by strongly disagree and strongly
agree, the three self-expression items were (1) “This product helps me to express
myself,” (2) “This product reflects my personality,” and (3) “This product enhances
my self-image.” The scale for the self-expressive value of the product was adopted
from Kim, Han, and Park’s (2001) study and modified to better fit within the con-
text of product category.
Among the product categories, business/dress shoes (M 5.40, SD 1.08)
and perfume (M 5.46, SD 1.40) were chosen as high self-expressive prod-
ucts, while disposable AA-size batteries (M 1.68, SD 1.01) and headache
medication (M 2.11, SD 1.31) were selected as low self-expressive prod-
ucts. These product categories were the most commonly listed products and
were not gender-specific products such as cosmetics, ties, or handbags.
The manipulation check showed that there were significant differences between
the self-expressive perceptions for each product. The mean differences
between the pairs of different product types [e.g., business/dress shoes and dis-
posable AA-size batteries (t 3.72, p 0.001)] were significant, whereas the
mean differences between the pairs of the same product types [e.g., business/dress
shoes and perfume (t 0.05, p 0.05)] were not significant. Therefore, the
manipulation check was successful.
Sample
The initial sample consisted of 152 undergraduate students (37.5% male and 62.5%
female) enrolled in introductory courses at a large southeastern university. Respon-
dents’ ages ranged from 18 to 25 years, with an average age of 20 years.
Participants who completed the survey were given extra course credit as an
incentive.
Subjects were randomly assigned to a questionnaire covering two of the four
product categories (i.e., business/dress shoes, perfume, disposable AA-size bat-
teries, and headache medication). Following Erdem and Swait’s (2004) approach,
five brands were presented in each product category to cover a wide range of mar-
ket share, which in turn led to increased potential heterogeneity in the brand
RESULTS
Assumption Check
Prior to the main analysis, several underlying assumptions for structural equa-
tion modeling (SEM) were checked. The fundamental statistical assumptions for
SEM analysis were similar to those for factor analysis: sampling adequacy, no
extreme multicollinearity, and normality (Hair et al., 1998). Kaiser-Meyer-
Olkin’s measure of sampling adequacy was 0.962, and Bartlett’s Test of Spheric-
ity index was significant ( p 0.001). Therefore, there was substantial evidence
for the planned factoring of the 20 items used in the study (Kaiser, 1974). Since
extracted communalities were 0.538 to 0.908 across all measurement items,
there was no extreme multicollinearity among the 20 measurement items.
The univariate normality assumption was satisfied because all skewness and
kurtosis values associated with each item were within the range of 1.96 (1.051
all skewness values 0.408; 1.138 all kurtosis values 0.166). The value
for multivariate normality was 1.683, which was well below the maximum cut-
off of the absolute value of 2.0 for multivariate normality (Kline, 2005).
Measurement Model
The hypothesized relationships were tested using a two-step procedure sug-
gested by Anderson and Gerbing (1988). For a two-step approach to structural
equation modeling, a confirmatory factor analysis (CFA) of the measurement
model was first conducted to evaluate whether the measurement items had the
appropriate properties to represent each construct. After achieving a satisfac-
tory fit in the measurement model, the structural model was estimated. Accord-
ing to Close et al. (2006), this method allows for rigorous testing of measurement
reliability and validity before subjecting the structural model to tests of fit
(p. 426). Since the normality assumption was met, the maximum likelihood esti-
mation method was used for the CFA in the study. Overall goodness-of-fit indices
were satisfactory [x2(155) 1315.73 ( p 0.001), GFI 0.92, AGFI 0.90,
RMSEA 0.072, NNFI 0.99, CFI 0.99, SRMR 0.031]. Since the meas-
urement model revealed a good fit, measurement respecification—a process of
adding or deleting estimated parameters from the original model (Hair et al.,
1988)—was not performed.
In this study, reliability and validity were evaluated using the pooled data
across four product categories. For internal reliability, Cronbach’s alpha coeffi-
cients were calculated for all items of each construct. Results indicated that all
the scales were considered to be reliable (Cronbach’s alphas for brand credibil-
ity 0.96, brand prestige 0.94, perceived quality 0.92, information costs
saved 0.88, perceived risk 0.84, and purchase intention 0.88).
1
The following brands were selected from the Mintel (2007) market research reports: business/dress
shoes: Bally, Kenneth Cole, Salvatore Ferragamo, Steve Madden, Rockport; perfume: Giorgio
Armani, Polo Ralph Lauren, Calvin Klein, Davidoff, Burberry; disposable AA-size batteries:
Duracell, Energizer, Eveready, Rayovac, Panasonic; headache medication: Advil, Bayer, Excedrin,
Motrin, Tylenol.
Factor
Constructs Measurement items Loadings
Brand 1. This brand delivers (or would deliver) what it promises. 0.88*
credibility 2. Product claims from this brand are believable. 0.90*
3. Over time, my experiences with this brand led me 0.90*
to expect it to keep its promises.
4. This brand is committed to delivering on its claims. 0.92*
5. This brand has a name you can trust. 0.88*
6. This brand has the ability to deliver what it promises. 0.91*
Brand 7. This brand is very prestigious. 0.94*
prestige 8. This brand has high status. 0.96*
9. This brand is very upscale. 0.87*
Perceived 10. The quality of this brand is very high. 0.93*
quality 11. In terms of overall quality, I would rate this brand as: 0.93*
Information 12. Knowing what I am going to get from this brand 0.82*
costs saved saves me time shopping around.
13. This brand gives me what I want, which saves 0.87*
me time and effort trying to do better.
14. I know I can count on this brand to be available 0.84*
in the future.
Perceived risk 15. I need a lot more information about this brand 0.85*
before I would buy it.
16. I never know how good this brand will be before 0.84*
I buy it.
17. To figure out what this brand is like, I would have 0.72*
to try it several times.
Purchase 18. I would never buy this brand. (R) 0.77*
intention 19. I would seriously consider purchasing this brand. 0.88*
20. How likely would you be to purchase this brand? 0.90*
Note: All items on 9-point “strongly disagree/strongly agree” scale were measured, except items 11
(9-point “low quality/high quality” scale), and 20 (“very unlikely/very likely” scale); (R) after an item indicates
that it was reversed for inclusion in the model; factor loading is based on standardized estimates.
* p 0.05.
Convergent validity was assessed by examining the factor loading for statis-
tical significance (Sujan, Weitz, & Kumar, 1994). As indicated in Table 1, all fac-
tor loadings were statistically significant (p 0.05) within an acceptable range
(from 0.72 to 0.96). In addition, the average variance extracted (AVE) was cal-
culated for rigorous testing of measurement validity. Fornell and Larcker (1981)
asserted that the average variance extracted (AVE) should be greater than the
recommended 0.50 to achieve convergent validity. As shown in Table 2, it was
found that the AVE values were greater than 0.50 for all constructs (0.616 all
AVE values 0.845), thus providing strong evidence of convergent validity.
Discriminant validity was evaluated by comparing the AVE estimates for
each construct with the square of the parameter estimates between the two
constructs. According to Fornell and Larcker (1981), discriminant validity is
achieved if the AVE of each construct exceeds the square of the standardized cor-
relations between the two constructs. All AVE estimates (e.g., 0.779 and 0.832
for brand credibility and brand prestige, respectively) were greater than the
Constructs AVE 1 2 3 4 5 6
Path
Paths Hypotheses Coefficients t-value
squared correlations between all constructs (e.g., 0.722 0.518). Thus, both
convergent validity and discriminant validity were established.
Multiple-Group Analyses
To test the proposed model for the moderating effect of self-expressive product
classification, a multigroup approach was employed, which requires “the use of
0.12*
Perceived Risk
* p 0.05.
x2 (159) 1417.47 (p 0.001); GFI 0.91; NFI 0.99; CFI 0.99; SRMR 0.034; RMSEA 0.073
the covariance matrix rather than the correlation matrix” (Maruyama, 1998,
p. 258). For the multiple-group analysis, the data were divided into separate
covariance matrices for high self-expressive product categories (N 660) and
low self-expressive product categories (N 840).
For comparison, the base model is defined as “the unconstrained model where
all paths are free to vary across both groups” (Garretson & Niedrich, 2004, p.
30). After the base model was run simultaneously, without invariance of path
coefficients, each gamma (i.e., all paths from exogenous variables to endoge-
nous variables) and beta path (i.e., all paths among endogenous variables) was
tested individually for equivalency by fixing each path coefficient in one group
to be equal to the other one by one.
Furthermore, a chi-square difference test was performed to examine the path
coefficient differences across groups (i.e., high vs. low self-expressive products).
Given that the chi-square difference test provides significant results, the path
coefficients were significantly different across groups. Therefore, it can be con-
cluded that there is a moderating role affecting the relationship between inde-
pendent and dependent variables (Kline, 2005).
Table 4 presents the path coefficients across groups (i.e., high vs. low self-
expressive products). There were significant differences [x2 (df 1) 3.84] in
path coefficients of brand credibility → perceived quality and brand prestige →
perceived quality and perceived risk → information costs saved between high
and low self-expressive product groups. The total effect comparison indicates that
brand credibility has more effect on purchase intention in the low self-expres-
sive product group (0.53) than in the high self-expressive product group (0.41),
whereas brand prestige has more influence on purchase intention in the high
self-expressive product group (0.40) than in the low self-expressive product
group (0.28).
Given the condition that the products were high self-expressive, H7 was not sup-
ported because the difference in coefficient size between brand credibility’s total
effect on purchase intention (0.41) and brand prestige’s total effect on purchase
intention (0.40) was very minimal (0.01). However, for low self-expressive
Standardized Path
Coefficients
* p 0.05.
product groups, brand credibility’s total impact (0.53) was stronger than brand pres-
tige’s total impact (0.28). Thus, H8 was supported. As anticipated, compared with
brand prestige, brand credibility was significantly more effective in influencing
brand purchase intention process when the product is low-self expressive.
DISCUSSION
The objective of this study was to investigate how brand credibility and brand
prestige affect consumer brand purchase processes and how brand credibility and
brand prestige work differently across multiple product categories that differ in
terms of their high and low self-expressive nature. Consistent with the results of
Erdem and Swait’s (1998) study, the findings of this research showed that brand
credibility positively influences perceived quality and information costs saved,
but negatively influences perceived risk, and all three latent constructs increase
brand purchase intention. More importantly, it was discovered that brand pres-
tige also exerts positive influences on purchase intention mediated by perceived
quality, information costs saved, and perceived risk. Despite the theoretical plau-
sibility, the linkages between brand prestige and its consequences have rarely
been studied in the marketing and consumer behavior literature. The results
Although the present study provides valuable insights into understanding the
combinatory mechanism of brand credibility and brand prestige across multiple
product categories, there are several limitations. First, the use of student sam-
ples may not be representative of the larger population. Another avenue for fur-
ther research is to replicate the proposed model on nonstudent samples to increase
the generalizability of the results. Second, the hypothesized relationships
for several brand constructs were tested with 20 different brands. However, this
study still relied on a limited number of product categories (i.e., business/dress
shoes, perfume, disposable AA-size batteries, and headache medication). There-
fore, future research is needed to examine the generalizability and robustness of
the proposed model with a larger set of product/service categories.
Finally, this study focused only on the United States. The validity of the pro-
posed model in regard to the combined effects of brand credibility and brand
prestige on purchase intention outside the United States is not clear. With the
growing trend in brand globalization, there is a need to apply the proposed model
across countries and/or cultures. This study suggests that two of Hofstede’s (1980)
cultural dimensions (i.e., uncertainty avoidance and power distance) may par-
ticularly affect the power of brand credibility and brand prestige in explaining
consumer brand choice because uncertainty avoidance is most clearly related to
brand credibility, but power distance is closely associated with brand prestige.
Erdem, Swait, and Valenzuela (2006) suggest that brand credibility’s impact on
consumer brand choice increases with uncertainty avoidance. On the other hand,
high power distance cultures tend to emphasize the importance of prestige in
vertical relationships between social classes (Hofstede, 1980). Thus, future
research should be a cross-country validation study that examines how brand cred-
ibility and brand prestige work differently under cultural differences, focusing
on the moderating effects of uncertainty avoidance and power distance.
REFERENCES
Aaker, D. A. (1991). Managing brand equity. New York: The Free Press.
Aaker, J. L. (1997). Dimensions of brand personality. Journal of Marketing Research, 34,
347–356.
Aaker, J. L. (1999). The malleable self: The role of self-expression in persuasion. Jour-
nal of Marketing Research, 36, 45–57.
Alden, D. L., Steenkamp, J. B. E. M., & Batra, R. (1999). Brand positioning through
advertising in Asia, North America, and Europe: The role of global consumer culture.
Journal of Marketing, 63, 75–87.
Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice:
A review and recommended two-step approach. Psychological Bulletin, 10, 441–423.
Batra, R., Ramaswamy, V., Alden, D. L., Steenkamp, J. B. E. M., & Ramachander, S.
(2000). Effects of brand local and nonlocal origin on consumer attitudes in developing
countries. Journal of Consumer Psychology, 9, 83–95.
Correspondence regarding this article should be sent to: Jooyoung Kim, Department of
Advertising and Public Relations, Grady College of Journalism and Mass Communica-
tion, The University of Georgia, Athens, GA 30602-3018 ( jykim@uga.edu).